Christian Terwiesch, a Wharton professor, was having dinner with his older children when the topic of artificial intelligence came up. Both of his children have been experimenting with emerging technologies in their own fields: “one is interested in design…and the other is interested in computer science.”
His son eventually inspired ChatGPT to describe a sorting algorithm “using Homer Simpson words.”
“Everything was humorous,” Terwiesch remarked, adding that “from a computer science standpoint, it was accurate, and it was so simple as a user interface.”
Users of ChatGPT may enter any prompt—The Inquirer previously requested it to write a piece about Gritty—and the programme can react almost instantaneously with a response written so clearly that it’s easy to think it was written by a person rather than a robot. Some examples demonstrate ChatGPT’s capacity to compose essays, legal documents, and even whole songs in the manner of an artist.
Terwiesch was eager to put it to the test. He gave ChatGPT questions from his operations management final exam—a topic on which he essentially authored the book—to see whether the technology could pass an MBA-level course.
That it did. Terwiesch gave the bot a “B to B-” score, but highlighted in his study report that some replies had “surprising blunders in quite easy arithmetic at the sixth-grade level.”
He claims that the mathematical flaws demonstrate that we are still a long way from adopting current technology to replace educated people.
“Consider a medical expert deciding what dose of a medicine to provide… This item would be absolute crap if you automated it as an investment adviser “Terwiesch said. In these instances, experts must “avoid this technology.”
Finding use for an experimental technology
Terwiesch, who teaches in the MBA programme at Wharton, urges colleagues to examine “opportunities where we might think about enhancing our learning experience” by incorporating A.I. technologies into the classroom. Preparing testing materials—Terwiesch requested it to develop new exam questions with some success, according to his research—and other approaches to alleviate instructors’ responsibilities across all grade levels, including primary and public schools, might be part of that.
“We all see and respect how all these professors come to work and work their buttocks off in a really challenging work environment,” Terwiesch added, pointing to cases when pupils have previously used ChatGPT to cheat on examinations, a source of concern for many detractors.
Terwiesch favours a complete ban on taking driver’s licence or CPA examinations since the aim of such exams is to confirm the test taker is competent to do certain abilities. However, the goal of teaching and learning is to “connect with the content,” not just memorise it.
Terwiesch believes that technology may be utilised for good in the classroom.
“Our responsibility as educators is to utilise technology to engage [students] in new ways,” Terwiesch said. “We have to create methods in the curriculum where the deepest skill that we truly want to teach is taught in a creative manner, because if we wind up in a position that is as good as before the introduction of ChatGPT, I believe we have wasted an amazing opportunity.”
He cited the example of employing the programme to act in the place of French philosophers, allowing students to question them in “real time.” (Yes, ChatGPT does speak French.) “How can we reinvent teaching French in a world when we have hundreds of French pen pals?” he wondered. Terwiesch cautioned that ChatGPT’s knowledge was not infinite. This is mostly due to the fact that it is based on “what it has seen in the past,” which means that the technology should be considered more as a support for improvements in domains where it may be valuable.
“We should not consider that this is the end of human thought and that robots are taking over,” Terwiesch added.
You might also be interested in reading, A novel method for making hydrogen peroxide without generating carbon dioxide has been devised
